PASDiff: Physics-Aware Semantic Guidance for Joint Real-world Low-Light Face Enhancement and Restoration
This work addresses the challenge of joint real-world low-light face enhancement and restoration for applications in computer vision, such as surveillance or photography, by introducing a novel method that integrates physical constraints and facial priors, though it is incremental in improving upon existing joint models.
The paper tackles the problem of enhancing and restoring face images captured in real-world low-light conditions, which suffer from degradations like low illumination, blur, and noise, by proposing PASDiff, a physics-aware semantic diffusion model that outperforms existing methods with a superior balance of natural illumination, color recovery, and identity consistency.
Face images captured in real-world low light suffer multiple degradations-low illumination, blur, noise, and low visibility, etc. Existing cascaded solutions often suffer from severe error accumulation, while generic joint models lack explicit facial priors and struggle to resolve clear face structures. In this paper, we propose PASDiff, a Physics-Aware Semantic Diffusion with a training-free manner. To achieve a plausible illumination and color distribution, we leverage inverse intensity weighting and Retinex theory to introduce photometric constraints, thereby reliably recovering visibility and natural chromaticity. To faithfully reconstruct facial details, our Style-Agnostic Structural Injection (SASI) extracts structures from an off-the-shelf facial prior while filtering out its intrinsic photometric biases, seamlessly harmonizing identity features with physical constraints. Furthermore, we construct WildDark-Face, a real-world benchmark of 700 low-light facial images with complex degradations. Extensive experiments demonstrate that PASDiff significantly outperforms existing methods, achieving a superior balance among natural illumination, color recovery, and identity consistency.